Humans can monitor their movements, an ability helpful to detect errors and improve future motor learning. To quantify this motor metacognitive ability, experimenters typically ask participants to make, and then introspect on, a voluntary movement. But voluntary movements include several processes like motor intentions, planning, and efferent commands; as well as afferent sensory and proprioceptive consequences. As a result, it is unclear whether motor metacognitive ability results from participants monitoring efferent or afferent processes (or both). To answer this question, we focussed on imagined movements, that have been shown to be functionally and computationally equivalent to executed movements. This allowed us to investigate efferent information, isolated from the associated proprioceptive and sensory feedback. Participants first imagined one of two movements, rated the vividness of the motor imagery, and subsequently executed a movement. We collected movement data and vividness ratings in two conditions that differed in whether the executed movement matched (congruent) or differed (incongruent) from the imagined movement. We reasoned that, if participants have metacognitive access to efferent information, the expected congruency effect (faster reaction and movement times on congruent than incongruent trials) would scale with reported vividness. We report the results of two experiments (n=30; n=45) that differed in the movements imagined and performed, where we indeed found the expected interaction effect, suggesting that young, healthy adults have weak access to efferent motor commands alone, independently from afferent information.